Varminterror
Member
- Joined
- Jul 17, 2016
- Messages
- 15,008
Linearity is an odd duck, for me. If I take 5 certified weights, place them one at a time on the scale, the precision should be +/- 0.001g.
Then when I place the weights on the scale together (in any combination which negates actual weight variability stacking), the readout might be 0.002 higher or lower than expected at different magnitudes.
In other words, you might not get a perfectly “linear” output if you add a linear progression of weights. It might read a little high at 30g, a little low at 60, dead on at 90, and a little low again at 120grams, within that 0.002g linearity spec. A 60g measured load might not actually be exactly half of a 120grn measured load.
If you weigh the same 50g weight (50.000, for sake of argument) 100 times, it should never show more than 50.001 or less than 49.999 if calibrated with a 50.000 check weight. If calibrated with a 100.000, you could consistently be reading 49.998, as high as 49.999, and as low as 49.997 (extreme end of the 0.002 linearity. But put two on the scale, and you might find a pair of weights which can’t exceed 50.001 each will weigh 100.004. Why? Because the scale may “drift” from 50 to 100, such it’s off by +/-0.002.
So the precision is the “repeatability.” Linearity is the margin for drift around non-calibration values. The the same weight would never read 50.002 and 49.998 (50+/-0.002), but it could read 49.997-49.999 (linearity error 0.002 down error from 50.000, then precision within +/-0.001gram).
Here’s a picture below. For the sake of argument, the calibration should lock the red curve to the blue line at the calibration weight points. Outside of those points, the linearity could bobble above or below “perfect linearity”. So what you know here: a measured 50grn load might not actually be perfectly 1/2 of a measured 100grn load - within 0.002g. But if you measure 100 50grn loads 100 times each, you know they will all be within 49.999 and 50.001.
The real take away, for me, is to calibrate with weights near my charge weights, and to use the same balance and same check weights for my load development and my bulk loading.
For reloading, however, it’s less critical. Do your load work up on the same scale, and calibrate with the same check weights, and you’d be falling in the same zone of linearity “error” for every batch loaded.
Then when I place the weights on the scale together (in any combination which negates actual weight variability stacking), the readout might be 0.002 higher or lower than expected at different magnitudes.
In other words, you might not get a perfectly “linear” output if you add a linear progression of weights. It might read a little high at 30g, a little low at 60, dead on at 90, and a little low again at 120grams, within that 0.002g linearity spec. A 60g measured load might not actually be exactly half of a 120grn measured load.
If you weigh the same 50g weight (50.000, for sake of argument) 100 times, it should never show more than 50.001 or less than 49.999 if calibrated with a 50.000 check weight. If calibrated with a 100.000, you could consistently be reading 49.998, as high as 49.999, and as low as 49.997 (extreme end of the 0.002 linearity. But put two on the scale, and you might find a pair of weights which can’t exceed 50.001 each will weigh 100.004. Why? Because the scale may “drift” from 50 to 100, such it’s off by +/-0.002.
So the precision is the “repeatability.” Linearity is the margin for drift around non-calibration values. The the same weight would never read 50.002 and 49.998 (50+/-0.002), but it could read 49.997-49.999 (linearity error 0.002 down error from 50.000, then precision within +/-0.001gram).
Here’s a picture below. For the sake of argument, the calibration should lock the red curve to the blue line at the calibration weight points. Outside of those points, the linearity could bobble above or below “perfect linearity”. So what you know here: a measured 50grn load might not actually be perfectly 1/2 of a measured 100grn load - within 0.002g. But if you measure 100 50grn loads 100 times each, you know they will all be within 49.999 and 50.001.
The real take away, for me, is to calibrate with weights near my charge weights, and to use the same balance and same check weights for my load development and my bulk loading.
For reloading, however, it’s less critical. Do your load work up on the same scale, and calibrate with the same check weights, and you’d be falling in the same zone of linearity “error” for every batch loaded.